Monitoring openvz resources using nagios and snmp

2010-10-11T15:13:27Z

MalloryScoggin:

== snmpd configuration ==
Debian Etch example: [http://www.research-service.com custom essay]
<pre>
apt-get install snmpd
</pre>

edit '''/etc/default/snmpd''' : remove ''-u snmp'' and replace ''127.0.0.1'' with your ip (ie : 207.46.250.119), Full'''/etc/default/snmpd''' example:
<pre>
export MIBDIRS=/usr/share/snmp/mibs
SNMPDRUN=yes
SNMPDOPTS='-Lsd -Lf /dev/null -I -smux -p /var/run/snmpd.pid 207.46.250.119'
TRAPDRUN=no
TRAPDOPTS='-Lsd -p /var/run/snmptrapd.pid'
</pre>

For Debian 4.x:
<pre>
export MIBDIRS=/usr/share/snmp/mibs
SNMPDRUN=yes
SNMPDOPTS='-Lsd -Lf /dev/null -I -smux -p /var/run/snmpd.pid'
TRAPDRUN=no
TRAPDOPTS='-Lsd -p /var/run/snmptrapd.pid'
</pre>

Create user(my_username) and add new mib. Password need a min. of 8 charactes. Username only characters:
<pre>
/etc/init.d/snmpd stop
echo rouser my_username priv >> /etc/snmp/snmpd.conf
echo "extend .1.3.6.1.4.1.2021.51 beancounters /bin/cat /proc/user_beancounters" >> /etc/snmp/snmpd.conf
echo "extend .1.3.6.1.4.1.2021.52 vzquota /bin/cat /proc/vz/vzquota" >> /etc/snmp/snmpd.conf
echo createUser my_username MD5 my_password DES >> /var/lib/snmp/snmpd.conf
/etc/init.d/snmpd start
</pre>

(Note that the createUser command goes into a separate file. On Centos5 this file is located in /var/net-snmp/snmpd.conf. Make sure you stop snmpd before putting the createUser command there!).

Testing snmp:
<pre>
snmpwalk -v 3 -u my_username -l authPriv -a MD5 -A my_password -x DES -X my_password $(hostname -i)
</pre>

Warning: the minimum pass phrase length is 8 characters.

== nagios configuration ==
=== example nagios configuration ===
add to configuration:
<pre>
define command {
command_name check_snmp_openvz_on_port
# command_line /usr/local/bin/check_snmp_openvz.sh $HOSTADDRESS$ PORT USER PASSWORD
command_line /usr/local/bin/check_snmp_openvz.sh $HOSTADDRESS$ $ARG1$ $ARG2$ $ARG3$
}
</pre>

<pre>
define host {
host_name openvz-server
alias Serwer Openvz
address 207.46.250.119
use generic-host
contact_groups admins
}
</pre>

<pre>
define service{
use generic-service
host_name openvz-server
service_description Virtual Machines Limits
check_command check_snmp_openvz_on_port!161!my_username!my_password
max_check_attempts 1
}

</pre>

=== nagios plugin ===
It is shell script:
<source lang="bash">
# cat /usr/local/bin/check_snmp_openvz.sh
#!/bin/bash
HOST=$1
PORT=$2
USER=$3
PASS=$4
export FILE=/tmp/$HOST.beancounters
RET=0

DATA_TMP=`snmpwalk -v 3 -u $USER -l authPriv -a MD5 -A $PASS -x DES -X $PASS $HOST:$PORT .1.3.6.1.4.1.2021.51.4`
if [ "$?" != "0" ]; then
echo "Unknown snmp error"
exit 1
fi

DATA=`echo "$DATA_TMP"| perl -ne '/"(.*)"/ ; print "$1\n" ;'`

if [ -f $FILE ]; then
echo "$DATA" | perl -n -e'
use Data::Dumper;
my $file=$ENV{"FILE"};
my $ret=0 ;
my $vid ;
my $resource ;
my $held ;
my $maxheld ;
my $barrier ;
my $limit ;
my $failcnt ;
my %beancounters ;
my %beancounters_old ;
while(<STDIN>){
my %vmachine;
if ( /\D*(\d+):.*/ ){ $vid=$1; $beancounters{$vid}=\%vmachine ; }
if ( /^[\W\d]+([a-z]+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+).*/ ) {
$resource=$1 ;
$held=$2 ;
$maxheld=$3 ;
$barrier=$4 ;
$limit=$5 ;
$failcnt=$6 ;
${beancounters{$vid}}{$resource}=[$held , $maxheld , $barrier , $limit ,$failcnt ];
if ( ($held > $barrier) && ($barrier != 0) ) {
print "WARNING: Limits on $vid: $resource held->$held , barrier->$barrier ( limit->$limit ) " ;
$ret=1;
}
}
}

# read and parse old data
open(MYINPUTFILE, "<$file");
while(<MYINPUTFILE>){
my %vmachine;
if ( /\D*(\d+):.*/ ){ $vid=$1; $beancounters_old{$vid}=\%vmachine ; }
if ( /^[\W\d]+([a-z]+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+).*/ ) {
$resource=$1 ;
$held=$2 ;
$maxheld=$3 ;
$barrier=$4 ;
$limit=$5 ;
$failcnt=$6 ;
${beancounters_old{$vid}}{$resource}=[$held , $maxheld , $barrier , $limit ,$failcnt ];
}
}

foreach my $vmachine_id (keys %beancounters) {
foreach my $resource (keys %{$beancounters{$vmachine_id}} ) {
if ( defined($beancounters{$vmachine_id}{$resource}[4]) && defined($beancounters_old{$vmachine_id}{$resource}[4]) ){
my $failcnt=$beancounters{$vmachine_id}{$resource}[4];
my $failcnt_old=$beancounters_old{$vmachine_id}{$resource}[4];
my $held=$beancounters{$vmachine_id}{$resource}[0];
my $maxheld=$beancounters{$vmachine_id}{$resource}[1];
my $barrier=$beancounters{$vmachine_id}{$resource}[2];
my $limit=$beancounters{$vmachine_id}{$resource}[3];
if ( $failcnt_old < $failcnt ){
print "CRITICAL: Incrased failcnt $vmachine_id: $resource from $failcnt_old to $failcnt (held->$held , maxheld->$maxheld , barrier->$barrier , limit->$limit ) " ;
$ret=2;
}
}
}

}

# if ($ret == 0 ) { print "Ok. \n" ; }
# print Dumper(%beancounters_old) ;
# print "\n";
exit($ret);
'

RET1=$?
fi

echo "$DATA" > $FILE
#####################################################################################
######### quota check
#####################################################################################

DATA=`snmpwalk -v 3 -u $USER -l authPriv -a MD5 -A $PASS -x DES -X $PASS $HOST:$PORT .1.3.6.1.4.1.2021.52.4 \
| perl -ne '/"(.*)"/ ; print "$1\n" ;'`

if [ "$?" != "0" ]; then
echo "Unknown snmp error"
exit 1
fi

echo "$DATA" | perl -n -e'
my $vid ;
my $ret=0 ;
while(<STDIN>){
my %vid;
if ( /\D*(\d+):.*/ ){ $vid=$1; }
if ( /\s*(\S+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+).*/ ){
$resource=$1 ;
$usage=$2 ;
$softlimit=$3 ;
$hardlimit=$4 ;
$time=$5 ;
$expire=$6 ;
if ( $usage >= $softlimit ){
print "WARNING: VZquota limit exceeded on $vid: $resource usage->$usage, softlimit->$softlimit, hardlimit->$hardlimit, time->$time, expire->$expire " ;
$ret=1;
}
}
}
exit($ret);
'
RET2=$?

#####################################################################################
########### return
#####################################################################################

if [ $RET1 -gt $RET2 ]; then
RET=$RET1
else
RET=$RET2
fi

if [ $RET = 0 ]; then
echo Ok.
fi
exit $RET
</source>

=== check_vzquota Without SNMP ===
<source lang="bash">
#!/bin/bash
RET=0
DATA=`echo;sudo /usr/sbin/vzlist -1 2>/dev/null | xargs -I {} bash -c "echo {}:;sudo /usr/sbin/vzquota stat {} | sed 's/\*//g'"`
if [ -z "$DATA" ]; then
VPS_err=$(sudo /usr/sbin/vzlist -1 2>&1 1>/dev/null)
if [ -n "$VPS_err" ] && [ "$VPS_err" == "Container(s) not found" ]; then
echo "OK - $VPS_err";
exit 0;
else
if [ -n "$VPS_err" ]; then
echo "UNKNOWN - Error: $VPS_err";
else
echo "UNKNOWN - VZquota stats are not readable or empty. Maybe it is only readable for root and this script should be called by sudo.";
fi
exit 3;
fi
fi

echo "$DATA" | perl -n -e'
my $vid ;
my $ret=0 ;
my $crit="";
my $warn="";
my $ok="";
while(<STDIN>){
my %vid;
if ( /^(\d+):.*/ ){ $vid=$1; }
if ( /\D*(\d+):.*/ ){ $vid=$1; }
if ( /\s*(\S+)\s+(\d+)\s+(\d+)\s+(\d+).*/ ){
$resource=$1 ;
$usage=$2 ;
$softlimit=$3 ;
$hardlimit=$4 ;
if ( $usage >= $hardlimit ){
$crit=$crit."VZquota limit exceeded on $vid: $resource usage->$usage, softlimit->$softlimit, hardlimit->$hardlimit, time->$time, expire->$expire " ;
$ret=2;
} elsif ( $usage >= $softlimit ){
$warn=$warn."VZquota limit exceeded on $vid: $resource usage->$usage, softlimit->$softlimit, hardlimit->$hardlimit, time->$time, expire->$expire " ;
$ret=1;
}
$ok=$ok."$vid:$resource $usage/$softlimit\n";
}
}
if ($ret == 0) {
print "OK - click on service-link for details...\n$ok";
} elsif ($ret == 1) {
print "WARNING - $warn\n";
} else {
print "CRITICAL - $crit\n";
}
exit($ret);
'
RET=$?
exit $RET
</source>
The script calls <code>/usr/sbin/vzlist</code> by sudo. When doing this it normally needs a password, which check_nrpe will not know. Because of this it is necessary that you append a line like the following to <code>/etc/sudors</code> (user name an path should be adapted to the right ones on your system):
nagios ALL=NOPASSWD: /usr/sbin/vzlist, /usr/sbin/vzquota

=== check_ubc Without SNMP ===
<source lang="bash">
#!/bin/bash
# Servicestate description can have a http-link to the openvz-wiki
# in case that a ressource is warning/critical. To use it:
# 1. set "escape_html_tags=0" in nagios/etc/cgi.cfg
# 2. set "my $linked=1;" in the first perl lines in this script
#
export FILE=/tmp/check_ubc
RET=0
ubc_file='/proc/user_beancounters';
DATA='';
if [ -r $ubc_file ]; then
DATA=`cat $ubc_file`
fi
if [ -z "$DATA" ]; then
echo "UNKNOWN - $ubc_file is not readable or empty. Maybe it is only readable for root and this script should be called by sudo.";
exit 3;
fi

if [ -f $FILE ]; then
echo "$DATA" | perl -n -e'
use Data::Dumper;
my $linked=1; # 0:plain text output, 1:resourcename is a http-link to OpenVZ-wiki
my $file=$ENV{"FILE"};
my $ret=0 ;
my $vid ;
my $resource ;
my $held ;
my $maxheld ;
my $barrier ;
my $limit ;
my $failcnt ;
my %beancounters ;
my %beancounters_old ;
while(<STDIN>){
my %vmachine;
if ( /\D*(\d+):.*/ ){ $vid=$1; $beancounters{$vid}=\%vmachine ; }
if ( /^[\W\d]+([a-z]+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+).*/ ) {
$resource=$1 ;
$held=$2 ;
$maxheld=$3 ;
$barrier=$4 ;
$limit=$5 ;
$failcnt=$6 ;
${beancounters{$vid}}{$resource}=[$held , $maxheld , $barrier , $limit ,$failcnt ];
if ( ($held > $barrier) && ($barrier != 0) ) {
print "WARNING: Limits on $vid: ".&url($resource,$linked)." held->$held , barrier->$barrier ( limit->$limit ) " ;
$ret=1;
}
#print "$vid:$resource $held Barrier:$barrier ";
}
}

# read and parse old data
open(MYINPUTFILE, "<$file");
while(<MYINPUTFILE>){
my %vmachine;
if ( /\D*(\d+):.*/ ){ $vid=$1; $beancounters_old{$vid}=\%vmachine ; }
if ( /^[\W\d]+([a-z]+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+).*/ ) {
$resource=$1 ;
$held=$2 ;
$maxheld=$3 ;
$barrier=$4 ;
$limit=$5 ;
$failcnt=$6 ;
${beancounters_old{$vid}}{$resource}=[$held , $maxheld , $barrier , $limit ,$failcnt ];
}
}

foreach my $vmachine_id (keys %beancounters) {
foreach my $resource (keys %{$beancounters{$vmachine_id}} ) {
if ( defined($beancounters{$vmachine_id}{$resource}[4]) && defined($beancounters_old{$vmachine_id}{$resource}[4]) ){
my $failcnt=$beancounters{$vmachine_id}{$resource}[4];
my $failcnt_old=$beancounters_old{$vmachine_id}{$resource}[4];
my $held=$beancounters{$vmachine_id}{$resource}[0];
my $maxheld=$beancounters{$vmachine_id}{$resource}[1];
my $barrier=$beancounters{$vmachine_id}{$resource}[2];
my $limit=$beancounters{$vmachine_id}{$resource}[3];
if ( $failcnt_old < $failcnt ){
print "CRITICAL: Incrased failcnt $vmachine_id: ".url($resource,$linked)." from $failcnt_old to $failcnt (held->$held , maxheld->$maxheld , barrier->$barrier , limit->$limit ) " ;
$ret=2;
}
#print "$vmachine_id: Old_Failcnt: $failcnt_old Failcnt: $failcnt \n";
}
}

}
sub url {
my ($name,$with_link) = @_;
if ($with_link) {
return "<a target=\"_blank\" href=\"http://wiki.openvz.org/".$name."#".$name."\">$name</a>";
} else {
return $name;
}
}
if ($ret == 0 ) { print "OK. \n" ; }
# print Dumper(%beancounters_old) ;
# print "\n";
exit($ret);
'

RET=$?
fi

echo "$DATA" > $FILE
exit $RET
</source>
The script needs to read the <code>/proc/user_beancounters</code> file. This is normally only readable for root. Because of this it is necessary that you append a line like the following to <code>/etc/sudors</code> (user name an path should be adapted to the right ones on your system):
nagios ALL=NOPASSWD: /usr/local/nagios/libexec/check_ubc

Also don't forget to consider this on your <code>nrpe.cfg</code>, so that you call the script with sudo:
command[check_ubc]=sudo /usr/local/nagios/libexec/check_ubc

OpenVZ Virtuozzo Containers Wiki - User contributions [en]

Monitoring openvz resources using nagios and snmp